Acute myocardial infarction (AMI), congestive heart failure (CHF), and pneumonia are three of the most common reasons for hospitalization in the Medicare program, and each is associated with high 30-day readmission rates. Although readmissions are common across all patient groups, black patients are more likely than white patients to be readmitted for reasons that are not well understood. The site of care likely plays an important role: care for minorities is concentrated among a small number of minority-serving hospitals (MSHs), which provide a lower quality of care and have worse outcomes, including higher readmission rates. However, even among MSHs, there are large variations: in preliminary analyses, we found three-fold variations in readmission rates (e.g. for AMI, 12.2% in the best versus 35.7% in the worst quartile) with some MSHs achieving results comparable to the best hospitals nationally. The Affordable Care Act (ACA) has made reducing readmissions a priority by requiring that Medicare lower payments to hospitals with high readmission rates and in preliminary analyses for this proposal, we found that MSHs have nearly three times greater odds of being in the worst quartile of performance nationally. If these hospitals fail to improve, the new policy is likely to worsen existing disparities in care. However, we know little about why some MSHs have low readmission rates while others lag behind. Helping MSHs reduce readmissions is critically important, but we lack the empirical data to make evidence-based recommendations. We propose to examine key patient-, hospital-, and market-level factors that explain the large variations in readmission among MSHs, and determine if the presence of these factors explains gaps in performance between MSHs and non-MSHs. We will use case studies to characterize readmission prevention practices at MSHs with the highest and lowest risk-adjusted readmission rates. We will then use a national survey of hospital Chief Medical Officers to create the first national picture of current knowledge, priorities, and programs around readmissions. Finally, we will use administrative databases to examine the degree to which these factors explain variations in readmissions among MSHs, whether the important factors we identify also explain variation at non-MSHs, and whether they explain disparities between non-MSHs and MSHs. Understanding the substantial variations in readmission rates among MSHs will allow us to identify key factors that are actionable and relevant to MSHs. These data will help policymakers craft effective and novel interventions help MSHs improve care for all their patients, thus reducing disparities in readmissions. PUBLIC HEALTH RELEVANCE: Acute myocardial infarction (AMI), congestive heart failure (CHF), and pneumonia are three of the most common reasons for hospitalization in the Medicare program, and each is associated with high 30-day readmission rates. Black patients have higher readmission rates than white patients for these conditions, and much of this difference is driven by site of care: minority-serving hospitals (MSHs) have higher rates of readmissions than non-MSHs. Even within MSHs, there are nearly three-fold differences in readmission rates, suggesting there are ample opportunities to improve. Our study will use a mixed-methods approach, including case studies, a national survey of hospital Chief Medical Officers, and multilevel modeling with Medicare data, to understand patient-, hospital-, and market-level factors that impact rates of readmissions at MSHs, with the goal of identifying actionable factors that could help policymakers craft effective policies to improve care among these providers and reduce racial disparities in this important health outcome.